Singlet Fission in Pentacene Dimers

Singlet fission in pentacene dimers

Johannes Zirzlmeiera, Dan Lehnherrb, Pedro B. Cotoc, Erin T. Chernickd, Rubén Casillasa, Bettina S. Basela, Michael Thossc, Rik R. Tykwinskid,1, and Dirk M. Guldia,1

aDepartment of Chemistry and Pharmacy & Interdisciplinary Center for Molecular Materials, Friedrich-Alexander-Universität Erlangen-Nürnberg, 91058 Erlangen, Germany; bDepartment of Chemistry, University of Alberta, Edmonton, AB, Canada T6G 2G2; cInstitute for Theoretical Physics & Interdisciplinary Center for Molecular Materials, Friedrich-Alexander-Universität Erlangen-Nürnberg, 91058 Erlangen, Germany; and dDepartment of Chemistry and Pharmacy & Interdisciplinary Center for Molecular Materials, Friedrich-Alexander-Universität Erlangen-Nürnberg, 91054 Erlangen, Germany

Edited by David R. Reichman, Columbia University, New York, NY, and accepted by the Editorial Board March 9, 2015 (received for review November 26, 2014)

Singlet fission (SF) has the potential to supersede the traditional excited states plus a certain amount of phonons are derived as solar energy conversion scheme by means of boosting the photon- the main relaxation products. In other words, the dissipated heat to-current conversion efficiencies beyond the 30% Shockley– leads to a decrease in SF yields and rates (15). Queisser limit. Here, we show unambiguous and compelling evi- In terms of exploiting SF for improving device performances, dence for unprecedented intramolecular SF within regioisomeric pen- e.g., hybrid solar cells, it is necessary to efficiently dissociate – tacene dimers in room-temperature solutions, with observed triplet correlated triplet pairs as they are formed, to overcome triplet quantum yields reaching as high as 156 ± 5%. Whereas previous triplet annihilation (5, 16). Rapid injection of electrons into studies have shown that the collision of a photoexcited chromophore fullerenes, perylene diimides, colloidal nanocrystals, semiconductor with a ground-state chromophore can give rise to SF, here we dem- substrates, etc. suggests a viable strategy. If successful, two charge onstrate that the proximity and sufficient coupling through bond or carriers might be produced per absorbed photon and the photo- space in pentacene dimers is enough to induce intramolecular SF currents of the device can reach external quantum efficiencies of where two triplets are generated on one molecule. more than 100% (5, 17, 18). A provocative debate has been ignited about the mechanism of SF at the molecular level (19, 20). Controversy exists around acene oligomers | excited states | singlet fission | multireference the electronic states that are involved in the process, the coupling perturbation theory | time-resolved spectroscopy among them, and the effective nuclear dynamics (14, 21–27). Two contrasting SF mechanisms have been traditionally postulated–– inglet fission (SF) is a spin-allowed process to convert one the direct and the two-step mechanism. These mechanisms differ Ssinglet excited state into two triplet excited states, namely a in the number and nature of the electronic states that are involved correlated triplet pair (1). The ability to effectively implement SF in the SF process. For the direct mechanism, the nonradiative processes in solar cells could allow for more efficient harvesting relaxation of the initially populated bright state proceeds via a of high-energy photons from the solar spectrum and allow for the correlated triplet pair state of singlet character––sometimes design of solar cells to circumvent the Shockley–Queisser perfor- called multiexcitonic (ME) state––which then dissociates into mance limit (2). Indeed, several recent studies have demonstrated two separated triplet excited states (28). For the two-step CHEMISTRY remarkably efficient solar cell devices based on SF (3–6). mechanism, the relaxation of the bright state occurs via an in- One requirement that needs to be met to achieve SF is that the termediate charge transfer (CT) state to the ME state. As in the photoexcited chromophore in its singlet excited state must share direct mechanism, the ME state eventually allows the two triplet its energy with a neighboring ground-state chromophore. As excited states involved to separate and undergo separate spin re- – such, the potential of coupled chromophores to afford two triplet laxation (11, 21 23). Recent works have, however, challenged excited states via SF has been elucidated in, for example, a tet- these traditional viewpoints on the SF mechanism and several new racene dimer with an SF yield of around 3% (3, 7). Additionally, models have been proposed. In particular, it has been suggested past experiments in single-crystal, polycrystalline, and amor- (13, 29, 30) that the initial excitation produces a coherent super- phous solids of pentacene have documented that the efficiency position of the lowest-lying absorbing state and the ME state, of SF relates to the electronic coupling between these two with the latter splitting into two separated triplet states after chromophores (8, 9). Hence, molecular ordering in terms of crystal packing, that is, proximity, distances, orbital overlap, etc., Significance is decisive with respect to gaining full control over and to fine- tuning interchromophoric interactions in the solid state (10, 11). In the present work, we show compelling evidence for the Of equal importance are the thermodynamic requirements, unprecedented intramolecular singlet fission at room temper- namely that the energy of the lowest-lying singlet absorbing state ature and in dilute solutions within a set of three different must match or exceed the energy of two triplet excited states regioisomeric pentacene dimers. Pump–probe experiments, ≥ (S1 2T1) (11). In light of both aspects, hydrocarbons such as which were complemented by theoretical calculations using –– –– acenes tetracene, pentacene, hexacene and their derivatives high-level ab initio multireference perturbation theory meth- are at the forefront of investigations toward a sound under- ods, corroborate triplet quantum yields as high as 156 ± 5%. standing and development of molecular building blocks for To this end, electronic couplings between the two pentacenes SF. In tetracenes, the singlet- and triplet-pair energy levels are = in the dimers, by virtue of through-bond or through-space in- nearly degenerate (S1 2T1), leaving no or little standard en- teractions, are decisive in tuning the rates of singlet fission. thalpy of reaction for SF (12). In solution, the latter is, however,

offset by sizable entropy rendering the process rather slow and, Author contributions: M.T., R.R.T., and D.M.G. designed research; J.Z., D.L., P.B.C., E.T.C., thus, inefficient (13). In addition, the low SF yield relates to the R.C., and B.S.B. performed research; J.Z., P.B.C., E.T.C., R.C., B.S.B., and D.M.G. analyzed dimer geometry. Its nature hinders electronic coupling through data; and M.T., R.R.T., and D.M.G. wrote the paper. space, leaving only through-bond coupling effective. The latter The authors declare no conflict of interest. is, however, insufficient to enhance the SF rate (7, 14). In stark This article is a PNAS Direct Submission. D.R.R. is a guest editor invited by the Editorial contrast, the relaxed triplet excited state in pentacenes has sig- Board. nificantly less than half the energy of the singlet excited state. In 1To whom correspondence may be addressed. Email: [email protected] or dirk.guldi@ turn, the thermodynamic SF requirement, that is (S1 ≥ 2T1), is fau.de. fulfilled for pentacenes rendering this process exothermic and This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10. unidirectional (13). Finally, from SF in hexacenes two triplet 1073/pnas.1422436112/-/DCSupplemental.

www.pnas.org/cgi/doi/10.1073/pnas.1422436112 PNAS | April 28, 2015 | vol. 112 | no. 17 | 5325–5330 Downloaded by guest on September 27, 2021 decoherence. On the other hand, recent theoretical works (14, 27) long wavelength absorption maxima in the 660–678-nm range in have proposed a model for SF in dimers, in which the ME state is toluene and 666–691-nm range in benzonitrile were determined, formed from the absorbing state via a superexchange mechanism together with short wavelength fluorescence maxima at 664 nm involving CT states, although the kinetics of the process is one- (m-2), 676 nm (o-2), and 670 nm (p-2) and fluorescence quantum step–like. All of the above suggests that the traditional classifica- yields ranging from 0.5% to 1.5% in toluene (Table 1). tion of SF mechanisms as direct or two-step is presumably too From the corresponding energetic differences we determined simple to describe the complexity of the process. lowest-lying singlet absorbing-state energies of 1.84 ± 0.05 eV for Until recently, most studies regarding SF have been carried the pentacene dimers o-2, m-2, and p-2. At first glance, all three out in the solid state (11). However, in a groundbreaking report, dimers give rise to absorption patterns that resemble those of Friend and coworkers showed that SF could be observed in so- 6,13-bis(triisopropylsilylethynyl)pentacene (TPc; see SI Appendix lution at room temperature for a pentacene derivative (31). Key for structure), including vibrational fine structure in toluene and to this discovery was the formation of an intermediate state via in benzonitrile in the range from 558 to 691 nm, although the the collision of a singlet excited-state pentacene and a second absorptions of the dimers are red-shifted relative to TPc. A closer pentacene that was in the ground state. This study broke for the look at the ortho-dimer o-2 suggests, however, strong electronic first time, to our knowledge, the dogma of molecular order as a couplings by means of through-space interactions in the ground mandate for SF and demonstrated that the order and packing state. This stems from the close spatial arrangement of the face-to- might not be as crucial as believed, leaving SF as an intrinsic face aligned pentacenes (Fig. 1) as seen in the optimized geom- property even in a state of disorder. etries (see SI Appendix for details). Finally, triplet excited-state In this present contribution, we report on intramolecular SF energies of 0.77 eV were derived for o-2, m-2,andp-2 from within a set of three different pentacene dimers (Fig. 1), reaching maxima at 1,610 nm observed in phosphorescence measurements triplet quantum yields as high as 156 ± 5%, as established by at 77 K (SI Appendix,Fig.S16). Thus, the singlet and triplet en- means of pump–probe experiments in solution. Experimental ergies of pentacene dimers 2 fulfill the thermodynamic re- results are complemented by theoretical calculations using high- quirement for SF, that is, twice the triplet excitation energy should level ab initio multireference perturbation theory methods. In the not exceed the lowest singlet excitation energy. Secondly, to elu- series of pentacene dimers, the two pentacenes are linked via a cidate a possible CT state, we probed m-2 by means of square phenylene spacer in an ortho-, meta-,andpara-arrangement to wave voltammetry. In benzonitrile (SI Appendix,Fig.S17), where impose geometrical control, which influences through-space and solvent stabilization is effective, we derived a first one-electron through-bond couplings in the ground- and excited state. The reduction of −1.43 V and a first one-electron oxidation of 0.44 V + electronic coupling element, which governs SF and triplet–triplet (versus Fc/Fc ). Neglecting any Coulombic attractive terms, the annihilation, is mediated through-bond in the linearly and cross- energy level of the intramolecular CT state for m-2 would have a conjugated para- and meta-isomers, respectively. In the ortho- lower limit of 1.87 eV and is, thus, quite achievable during exci- isomer, through-space coupling should dominate due to the unique tation. Please note that this intramolecular CT state places the spatial proximity of the pentacenes (32, 33). Through experiments, one-electron reduced form on one pentacene and the one-elec- we establish a sound picture of the SF mechanism in covalently tron oxidized form on the other pentacene, lowering attractive linked pentacene dimers, showing the involvement of CT states. interactions and facilitating its formation.

Results and Discussion Transient Absorption Spectroscopy and SF Kinetics. As a reference, − Synthesis. Synthesis of the pentacene dimers was based on TPc in Me-THF (3.0 × 10 4 M) was examined in pump–probe desymmetrization of 6,13-pentacenequinone to give building block experiments with an excitation wavelength of 656 nm to stimulate 1 (Fig. 1) (34). Dimers o-2, m-2,andp-2 were then assembled to population of only vibrational states of the first singlet excited state probe the effects of the geometry about the phenylene π-spacer. (SI Appendix,Fig.S18). Additionally, the photon flux was adjusted Sonogashira coupling of 1 with ortho-, meta-, and para-diiodo- to 6.6 × 109 photons per pulse to excite, on average, less than 10% benzene afforded intermediates 3–5, respectively, in ∼70% yield. of the ground-state TPc to rule out multiple excitations of a single SnCl2-mediated reductive aromatization in the presence of molecule. The major deactivation pathway for photoexcited TPc ± aq. H2SO4 yielded finally dimers o-2, m-2,andp-2.Theortho- includes a 12.3 0.1-ns intersystem crossing of the singlet excited phenylene–linked pentacene dimer o-2 has reasonable solubility in state to afford the corresponding triplet manifold. The corre- solvents such as CH2Cl2, CHCl3, and THF, whereas the corresponding triplet excited state deactivates with a rather long lifetime sponding meta-andpara-derivatives, m-2 and p-2 respectively, of 24.0 ± 0.5 μs to reinstate the singlet ground state. In terms of the have poor solubility in these solvents. singlet excited-state characteristics, maxima at 447, 508, 538, 572, 845, and 1,400 nm were noted, as well as minima at 595, 641, and Excited-State Energetics. Steady-state spectroscopies were used to 710 nm. Notably, the feature at 710 nm corresponds to stimulated examine the singlet- (ESinglet) and triplet-state energetics (ETriplet) emission. The triplet excited state has maxima at 465 and 498 nm of the pentacene dimers (SI Appendix, Figs. S14 and S15). First, and minima at 546, 587, and 641 nm in the differential absorption

Table 1. Spectroscopic data of pentacene dimers

−1 −1 −1 Pentacene dimer Solvent ESinglet [eV] ETriplet [eV] ΦF [%] kCT [s ] kSF [s ] kTTA [s ] ΦTriplet [%]

m-2 Toluene 1.88 0.77* 0.6 ± 0.1 3.97 ± 0.80 × 1010 1.11 ± 0.10 × 1010 3.85 ± 0.15 × 108 125 ± 5† Benzonitrile 1.86 0.4 ± 0.1 6.10 ± 0.70 × 1010 1.59 ± 0.15 × 1010 4.55 ± 0.15 × 108 156 ± 5† o-2 Toluene 1.87 — 1.6 ± 0.2 ———— ‡ Benzonitrile 1.85 —— 2.00 ± 0.80 × 1012 8.33 ± 0.20 × 1010 ‡ p-2 Toluene 1.83 — 0.3 ± 0.1 ———— ‡ Benzonitrile 1.79 —— 3.70 ± 1.50 × 1011 5.78 ± 0.50 × 1010 130 ± 10

Singlet excited-state energy (ESinglet), triplet excited-state energy (ETriplet), fluorescence quantum yields (ΦF), rate constants for charge transfer (kCT), rate constants for triplet growth (kSF), rate constants for triplet decay (kTTA), triplet excited-state quantum yields (ΦTriplet)form-2, o-2, and p-2 in toluene and benzonitrile. *Determined in frozen Me-THF at 77 K. † Determined by using the power-dependent method. ‡ The short-lived nature of the triplet excited states in o-2 hampers a meaningful triplet excited-state quantum yield determination.

5326 | www.pnas.org/cgi/doi/10.1073/pnas.1422436112 Zirzlmeier et al. Downloaded by guest on September 27, 2021 CHEMISTRY

Fig. 1. (Upper) Synthesis of pentacene dimers o-2, m-2, and p-2.(Lower) SF mechanism.

spectra. All of these observations are in excellent agreement with supra. The meta-isomer (m-2) was examined in benzonitrile and previously reported singlet and triplet excited-state absorptions at time delays as short as 0.5 ps singlet excited-state features are of TPc in solution (31). When using singlet molar extinction co- discernible in the form of maxima at 456, 510, 582, and 670 nm. 4 –1 –1 efficients of 8.3 × 10 M ·cm at the 447-nm maximum, in In contrast with the slow intersystem crossing seen for TPc, these combination with triplet molar extinction coefficients of 1.45 × 5 –1· –1 singlet excited-state features are rapidly replaced by strong 10 M cm at the 498-nm maximum, the computed TPc triplet absorptions, which are concentration independent in the range quantum yield is 16%. For this analysis, ground-state extinction −5 −4 – – from 1.0 × 10 to 1.0 × 10 M and, thus, strictly unimolecular. coefficients of 3.3 × 104 M 1·cm 1 at the 641-nm minimum and – – 1.6 × 104 M 1·cm 1 at the 587-nm minimum were used. These strong absorptions, which appear at time delays of around Next, the three pentacene regioisomers o-2, m-2, and p-2 were 200 ps, bear great resemblance with that of the triplet excited examined in solvents of varied polarity (i.e., toluene, THF, and state of m-2. In particular, maxima at 477, 510, 855, and 972 are benzonitrile) with photoexcitation at either 610 or 656 nm as complemented by minima at 610 and 664 nm and are a perfect shown in Fig. 2 as well as SI Appendix, Figs. S19, and S20, re- match to the photosensitized triplet excited-state spectra of the spectively, and the same low photon flux described for TPc––vide dimer m-2 (vide infra).

Zirzlmeier et al. PNAS | April 28, 2015 | vol. 112 | no. 17 | 5327 Downloaded by guest on September 27, 2021 Fig. 2. SF in pentacene dimer m-2. (Upper Left) Chemical structure. (Upper Right) Differential absorption spectra (visible and near-infrared) of the spectra shown in the lower right with time delays of 2.0 ps (black) and 190.0 ps (gray). (Lower Left) Time absorption profiles of the spectra shown in the lower right at 456 nm (black), 510 nm (red), 582 nm (orange), and 664 nm (gray) illustrating the dynamics of the singlet excited-state formation followed by the singlet to triplet transformation in the form of SF, and triplet–triplet annihilation. (Lower Right) Differential absorption changes (visible and near-infrared) obtained upon femtosecond pump–probe experi- − ments (610 nm) of m-2 (3.5 × 10 5 M) in argon-saturated benzonitrile at room temperature with several time delays between 0 and 5,500 ps.

A particularly striking effect observed for m-2 during the ac- fingerprints for the singlet and triplet excited states, respectively, celerated triplet excited-state formation is the long wavelength evolve in pump–probe experiments. Here, in contrast with p-2 and blue shift, from 670 to 664 nm, that is accompanied by a 155 ± m-2, strong through-space couplings between the two pentacene 10% intensification of the transient bleaching. This effect is moieties affect the triplet excited-state kinetics––both in terms notable upon 610-nm excitation rather than 656-nm excitation. If of formation and decay––and deconvolution results in upper the ground-state absorption depletion in this spectral region values of 0.5 ± 0.2 ps and 12.0 ± 0.3 ps, for formation and decay, were quantitative for the singlet excited state, this infers more respectively. The short-lived nature of the triplet excited states than one triplet excited state is formed per dimer. Nevertheless, in p-2 and o-2 hampers a meaningful triplet excited-state the triplet excited state is surprisingly short-lived and decays with quantum yield determination. The latter is corroborated in the concentration-independent kinetics. The only plausible modus geometry-optimized structures (SI Appendix,Fig.S27), which operandi is based on intramolecular triplet–triplet annihilation reveal, on one hand, the coplanarity between the two penta- governed by the unique scenario that more than one triplet ex- cenes in p-2 and m-2, and, on the other hand, only sizable cited state is localized on each dimer. Multiwavelength analyses through-space couplings in o-2. attest that the triplet excited-state formation is biexponential in benzonitrile, and it features two distinct steps with CT (τCT) and Triplet Quantum Yield and SF Yield Determination. Using the dif- SF (τSF) lifetimes of 16.4 ± 1.9 and 63.0 ± 6.3 ps, respectively. ferential absorption changes at 610 and 664 nm for m-2, for From the study of solvent dependence vide infra, we postulate which ground-state extinction coefficients of 2.8 × 104 and 5.3 × – – the involvement of an additional state that possesses an intra- 104 M 1·cm 1 have been determined, the singlet and triplet mo- molecular CT character and mediates the transition to the ME lar extinction coefficients are calculated as 7.9 × 104 and 1.4 × – – state (either directly or virtually). A closer look at the transient 105 M 1·cm 1, respectively. In recent studies, it was hypothesized absorption spectra taken at delay times of 80–300 ps reveals a that the bleaching at 610 nm is elusive for the triplet excited 445-nm shoulder, which is likely to stem from the CT state (SI state, whereas that the bleaching at 664 nm serves as an internal Appendix, Fig. S21). Once formed, the triplet excited-state decay standard to calibrate the singlet excited state (31). Because the is monoexponential for m-2, with a rather short triplet–triplet meta-dimer m-2 has only very weak ground-state absorption at annihilation lifetime (τTTA) of 2.2 ± 0.1 ns. 456 nm, the observed amplitude of differential absorption at 456 Turning our attention to the para-andortho-isomers (p-2 and o- nm and 0.5 ps relates entirely to the singlet excited-state con- 2), their singlet excited-state features are discernible immediately centration formed after excitation. In doing so, differential ab- after photoexcitation in benzonitrile. For p-2, these are maxima at sorption changes of 0.028 a.u. at 456 nm and 0.0726 a.u. at 462 and 1,117 nm and a minimum at 686 nm (SI Appendix,Fig. 510 nm were converted into a triplet quantum yield (ΦTriplet)of S19). The features of the singlet state of p-2, however, transform 145 ± 10%. This high triplet quantum yield along with its ul- even faster than in m-2 into the triplet excited-state maxima at 475 trafast formation provides strong evidence that the triplet excited and 508 nm as well as its minimum at 670 nm. From multiwave- state for m-2 forms by intramolecular SF between the two penta- length analyses we derive lifetimes for the growth and decay of the cenes. The triplet quantum yield for p-2 was determined by ex- triplet excited state as 2.7 ± 1.0 and 17.3 ± 1.3 ps, respectively. It is trapolation to time zero as 130%. likely that the dramatic difference in behavior in p-2 versus m-2 Taking the ratio from 510/456 nm as an indicator for the re- kinetics derives from enhanced through-bond electronic coupling spective quantum yield, we note that it increases from toluene in the ground state of the para-isomer as a means to impact (126 ± 3%) to THF (132 ± 2%) and to benzonitrile (145 ± 10%). both SF and triplet–triplet annihilation (32, 33). As such, the elec- We surmise from this trend that the mediating step, which dictates tronic coupling matrix elements as the decisive factor to govern the triplet excited-state formation, features CT-type character and CT are larger in p-2 than in m-2 (32). For o-2, the 472- and 507-nm this operates through-bond for dimer m-2. Such a postulate fits well

5328 | www.pnas.org/cgi/doi/10.1073/pnas.1422436112 Zirzlmeier et al. Downloaded by guest on September 27, 2021 with the fact that the triplet excited-state kinetic growth is biex- growth are pentacene concentration dependent. Complemen- ponential and that the kinetics are solvent dependent. Biexpo- tary analyses of the N-MFP triplet excited-state decay as well as nential fits of these data, for example, yields lifetimes of 25.2 ± of the pentacene triplet excited-state growth provide the means 4.8/90.2 ± 9.0 ps in toluene, 18.6 ± 2.9/70.3 ± 7.0 ps in THF, and to derive the second-order energy transfer rate constants (SI 16.4 ± 1.9/63.0 ± 6.3 ps in benzonitrile. In contrast, the intramolecular Appendix, Fig. S24). The rate constants in deoxygenated toluene – – – – triplet–triplet annihilation lacks any appreciable solvent polarity are 1.6 ± 0.1 × 109 M 1·s 1 for o-2, 3.4 ± 0.3 × 109 M 1·s 1 for – – dependence as evidenced by nearly constant lifetimes of 2.6 ± 0.1 ns m-2, and 8.8 ± 0.5 × 109 M 1·s 1 for p-2, which infers that diffusion in toluene, 2.5 ± 1.0 ns in THF, and 2.2 ± 0.1 ns in benzonitrile. controls the transduction of triplet excited-state energy. Once Alternatively, the power-dependent method using the differ- formed, the pentacene triplet excited states in o-2, m-2,andp-2 ential absorption changes at 670 and 664 nm was used to calculate are subject to concentration-dependent deactivations via effi- the transient extinction coefficient of the bleaching as 5.7 × 104 cient intermolecular triplet–triplet annihilation with lifetimes in – – and 8.3 × 104 M 1·cm 1 for the singlet and the triplet excited state, the range of tens of microseconds. Intramolecular contributions respectively. Consequently, we obtained a triplet quantum yield under these conditions play no major role. A rate constant of 1.5 ± – – for the meta-dimer (m-2)of156± 5%. The main advantage of this 0.5 × 108 M 1·s 1 was derived for m-2 from pseudo–first-order method is that it allows the determination of transient extinction analyses (SI Appendix,Fig.S25). From Figs. 2 (i.e., direct excita- coefficients correlated by the amount of excited molecules in the tion experiments) and 3 (i.e., triplet sensitization experiments) the exact range of excitation without using an external reference (31). singlet molar extinction coefficient at 456 nm, the triplet molar To complement the power-dependent studies, we also gener- extinction coefficient at 506 nm, and the triplet quantum yield – – – – ated the triplet excited-state signatures of all three dimers by were determined for m-2 as 7.4 × 104 M 1·cm 1,1.5× 105 M 1·cm 1, means of triplet sensitization experiments in toluene. In these and 130 ± 10%, respectively (see SI Appendix for details). experiments, an N-methylfulleropyrrolidine (N-MFP; see SI Appendix for structure) with a triplet quantum yield close to unity Theory. To characterize the electronic states involved in the SF and a triplet excited-state energy of 1.5 eV was taken and process, we have used the second-order complete active space selectively excited at 480 nm, which coincides with a minimum in perturbation theory with a complete active space self-consistent − the pentacene ground-state absorption. N-MFP (8.0 × 10 5 M) field reference wavefunction (CASPT2/CASSCF) approach (see was photoexcited, leading to the sequential formation of its SI Appendix for details). This ab initio multireference perturba- short-lived singlet and its long-lived triplet excited states with tion theory method is known to provide accurate results for their 695- and 900-nm fingerprint absorptions, respectively. In electronic states of very different chemical character like those the presence of the three pentacene regioisomers o-2, m-2, and involved in SF (35). The results, depicted in SI Appendix, Table − p-2 as well as TPc in concentration regimes from 1.0 × 10 5 to 1.0 × S1, reveal a set of closely lying electronic excited states of ME − 10 4 M, a set of newly developing transient absorption features and locally excited (LE) character. Whereas in m-2 and p-2 the evolves with the depletion of the N-MFP triplet excited state first excited state, S1, is of ME character and S2 and S3 are of LE (Fig. 3 as well as SI Appendix, Figs. S22 and S23). The meta- character, respectively, this order changes in o-2. In all cases, dimer m-2, for example, shows a transient maximum at 506 nm and however, the ME state is quasi-degenerate with the initially ex- minima at 559, 605, and 661 nm. While for TPc pentacene, a cited LE states and, therefore, a direct pathway for SF is ener-

503-nm maximum as well as 547-, 592-, and 643-nm minima were getically possible. The results obtained in vacuo predict CT states CHEMISTRY noted, for o-2 and p-2, 512/507-nm maxima as well as 570-, 614-, located at significantly higher energies than the LE and ME 664-nm and 562-, 614-, and 668-nm minima, respectively, were states. Additional calculations using the polarization continuum recorded. Importantly, a closer look reveals that the decay and model (see SI Appendix for details) for benzonitrile reveal a

Fig. 3. Triplet–triplet sensitization of pentacene dimer m-2.(Upper Left) Chemical structure. (Upper Right) Differential absorption spectrum (visible and near- infrared) of the spectra shown in the lower right with a time delay of 6.8 μs represent- ing the triplet excited state of m-2.(Lower Left) Time absorption profiles of the spectra shown in the lower right at 506 nm (black), 605 nm (red), 661 nm (orange), and 697 nm (gray) illustrating the dynamics of the N-MFP singlet excited-state formation followed by the intersystem crossing to the corresponding N-MFP triplet excited state and transduction of triplet excited- state energy to m-2.(Lower Right)Differ- ential absorption changes (visible and near-infrared) obtained upon femtosecond pump–probe experiments (480 nm) of N-MFP − − (8.0 × 10 5 M) and m-2 (1.0 × 10 4 M) in argon- saturated toluene at room temperature with several time delays between 0 and 400 μs.

Zirzlmeier et al. PNAS | April 28, 2015 | vol. 112 | no. 17 | 5329 Downloaded by guest on September 27, 2021 pronounced stabilization of the CT states which locates them pentacene dimers are rapidly and efficiently converted to triplet ∼0.35 eV above the lowest-lying LE and ME states in m-2 (SI excited states via SF, this is considered only as a preliminary step Appendix). Taking into account the experimentally observed toward realizing SF enhanced devices. In this context, a sound solvent dependence of the triplet yield and the kinetics, our re- platform en route toward charge separation with suitable electron sults indicate that CT states participate in SF either as virtual acceptors and charge collection at suitable electrodes is forming. states via a superexchange mechanism (14) or directly if nuclear relaxation of the lowest-lying absorbing state facilitates pop- Materials and Methods ulation of the CT state via internal conversion. In addition to The synthesis of o-2, m-2, and p-2, materials, general methods, and com- the characterization of the excited singlet states, we also have putational details are described in SI Appendix. characterized the excited triplet states of m-2. The results in SI For the photophysical characterization (see SI Appendix for details), the Appendix, Table S3 reveal a strong T1 → T9 excitation at 2.43 eV, samples were placed in fluorometric cuvettes with different pathways and, which corresponds to the pronounced triplet feature at 510 nm when necessary, purged with argon. Steady-state UV-vis absorption spectra observed in the transient absorption spectra (Fig. 3). were acquired at room temperature (RT) using a Perkin Elmer Lambda 2 spectrometer. Steady-state fluorescence spectra were carried out at a FluoroMax3 Conclusions spectrometer from Horiba in the visible detection range (RT) and at a FluoroLog3 In conclusion, we used pump–probe experiments and comple- spectrometer from Horiba in the near infrared. Femtosecond (fs) transient ab- mentary theoretical calculations using multireference perturbation sorption experiments were carried out with an amplified Ti:Sapphire CPA-2110 fs theory to corroborate for the first time, to our knowledge, in- laser system (Clark MXR: output 775 nm, 1 kHz, 150-fs pulse width) using transient tramolecular SF in pentacene dimers. Three regioisomeric absorption pump/probe detection systems (Helios and Eos, Ultrafast Systems). pentacene dimers, in which the two pentacenes are linked via a Details of the calculations can be found in the SI Appendix. Briefly, the ––o-2 m-2 geometries of o-2, m-2, and p-2 were optimized using density functional phenylene spacer in ortho-, meta,orpara-position , , and − − p-2––feature triplet quantum yields as high as 156 ± 5% medi- theory. In all systems, i-Bu3Si groups have been modelled using H3Si to ensure computational tractability. For the different systems investigated, ated either by through-space or through-bond electronic cou- vertical excitation energies were calculated at the CASPT2/CASSCF/ANO-S- plings in the ground- and excited state. As recently hypothesized, VDZP level of theory (35, 36). In all calculations, an active space of 8 electrons mixing between the singlet excited state and the CT state might in 8 orbitals (comprising the HOMO-1, HOMO, LUMO, and LUMO+1 orbitals be helpful in terms of overcoming the activation barrier (14). It is per pentacene unit; see SI Appendix, Fig. S26) was used. in m-2 where the right balance between triplet formation and triplet decay optimizes the SF yield. Such an efficient SF reaches ACKNOWLEDGMENTS. Generous allocation of computing time at the com- well beyond the 3% mark found in tetracene dimers, for which, puting centers Erlangen (Regionales Rechenzentrum Erlangen), Munich incidentally, it is the para-isomer that revealed the highest SF (Leibniz-Rechenzentrum der Bayerischen Akademie der Wissenschaften in yields (7). The demonstrated depletion of the ground state, the München), and Jülich (Jülich Supercomputing Centre) is gratefully acknowl- quantitative analysis of forming more than one triplet excited edged. Funding is gratefully acknowledged from the Emerging Fields initiative “Singlet Fission” supported by Friedrich-Alexander-Universität Erlangen-Nürn- state per singlet, and the very low concentrations used corrob- berg, as well as the Cluster of Excellence Engineering of Advanced Materials orate the intramolecular nature of SF. Although our studies and “Solar Technologies Go Hybrid”––an initiative of the Bavarian State Min- demonstrated that the singlet excited states in all three different istry for Science, Research, and Art.

1. Smith MB, Michl J (2010) Singlet fission. Chem Rev 110(11):6891–6936. 19. Zimmerman PM, Musgrave CB, Head-Gordon M (2013) A correlated electron view of 2. Shockley W, Queisser HJ (1961) Detailed balance limit of efficiency of p-n junction solar singlet fission. Acc Chem Res 46(6):1339–1347. cells. JApplPhys32(3):510–519. 20. Johnson JC, Nozik AJ, Michl J (2013) The role of chromophore coupling in singlet 3. Ehrler B, Musselman KP, Böhm ML, Friend RH, Greenham NC (2012) Hybrid pentacene/a- fission. Acc Chem Res 46(6):1290–1299. silicon solar cells utilizing multiple carrier generation via singlet exciton fission. Appl 21. Feng X, Luzanov AV, Krylov AI (2013) Fission of entangled spins: An electronic Phys Lett 101(15):153507. structure perspective. J Phys Chem Lett 4(22):3845–3852. 4. Jadhav PJ, et al. (2012) Triplet exciton dissociation in singlet exciton fission photo- 22. Parker SM, Seideman T, Ratner MA, Shiozaki T (2014) Model Hamiltonian analysis of voltaics. Adv Mater 24(46):6169–6174. singlet fission from first principles. J Phys Chem C 118(24):12700–12705. 5. Lee J, Jadhav P, Baldo MA (2009) High efficiency organic multilayer photodetectors 23. Piland GB, Burdett JJ, Dillon RJ, Bardeen CJ (2014) Singlet fission: From coherences to – based on singlet exciton fission. Appl Phys Lett 95(3):033301. kinetics. J Phys Chem Lett 5(13):2312 2319. 6. Reusswig PD, Congreve DN, Thompson NJ, Baldo MA (2012) Enhanced external 24. Teichen PE, Eaves JD (2012) A microscopic model of singlet fission. J Phys Chem B – quantum efficiency in an organic photovoltaic cell via singlet fission exciton sensitizer. 116(37):11473 11481. Appl Phys Lett 101(11):113304. 25. Zeng T, Hoffmann R, Ananth N (2014) The low-lying electronic states of pentacene – 7. Müller AM, Avlasevich YS, Schoeller WW, Müllen K, Bardeen CJ (2007) Exciton fission and their roles in singlet fission. J Am Chem Soc 136(15):5755 5764. and fusion in bis(tetracene) molecules with different covalent linker structures. JAm 26. Zimmerman PM, Bell F, Casanova D, Head-Gordon M (2011) Mechanism for singlet fission in pentacene and tetracene: From single exciton to two triplets. J Am Chem Soc Chem Soc 129(46):14240–14250. 133(49):19944–19952. 8. Johnson JC, Reilly TH, III, Kanarr AC, van de Lagemaat J (2009) The ultrafast photo- 27. Berkelbach TC, Hybertsen MS, Reichman DR (2013) Microscopic theory of singlet ex- physics of pentacene coupled to surface plasmon active nanohole films. J Phys Chem C citon fission. I. General formulation. J Chem Phys 138(11):114102. 113(16):6871–6877. 28. Chan WL, et al. (2011) Observing the multiexciton state in singlet fission and ensuing 9. Marciniak H, Pugliesi I, Nickel B, Lochbrunner S (2009) Ultrafast singlet and triplet ultrafast multielectron transfer. Science 334(6062):1541–1545. dynamics in microcrystalline pentacene films. Phys Rev B 79(23):235318. 29. Chan WL, et al. (2013) The quantum coherent mechanism for singlet fission: Experi- 10. Wilson MW, et al. (2011) Ultrafast dynamics of exciton fission in polycrystalline ment and theory. Acc Chem Res 46(6):1321–1329. pentacene. J Am Chem Soc 133(31):11830–11833. 30. Akimov AV, Prezhdo OV (2014) Nonadiabatic dynamics of charge transfer and singlet 11. Yost SR, et al. (2014) A transferable model for singlet-fission kinetics. Nat Chem 6(6): fission at the pentacene/C interface. J Am Chem Soc 136(4):1599–1608. – 60 492 497. 31. Walker BJ, Musser AJ, Beljonne D, Friend RH (2013) Singlet exciton fission in solution. 12. Bayliss SL, et al. (2014) Geminate and nongeminate recombination of triplet excitons Nat Chem 5(12):1019–1024. formed by singlet fission. Phys Rev Lett 112(23):238701. 32. Gaab KM, Thompson AL, Xu J, Martínez TJ, Bardeen CJ (2003) Meta-conjugation and 13. Chan WL, Ligges M, Zhu XY (2012) The energy barrier in singlet fission can be over- excited-state coupling in phenylacetylene dendrimers. J Am Chem Soc 125(31):9288–9289. – come through coherent coupling and entropic gain. Nat Chem 4(10):840 845. 33. Quardokus RC, et al. (2012) Through-bond versus through-space coupling in mixed- 14. Berkelbach TC, Hybertsen MS, Reichman DR (2013) Microscopic theory of singlet ex- valence molecules: Observation of electron localization at the single-molecule scale. citon fission. II. Application to pentacene dimers and the role of superexchange. J Am Chem Soc 134(3):1710–1714. J Chem Phys 138(11):114103. 34. Lehnherr D, Murray AH, McDonald R, Tykwinski RR (2010) A modular synthetic approach 15. Busby E, et al. (2014) Multiphonon relaxation slows singlet fission in crystalline hex- to conjugated pentacene di-, tri-, and tetramers. Angew Chem Int Ed 49(35):6190–6194. acene. J Am Chem Soc 136(30):10654–10660. 35. Roos BO, Fülscher MP, Malmqvist P-Å, Merchán M, Serrano-Andrés L (1996) Multi- 16. Rao A, et al. (2010) Exciton fission and charge generation via triplet excitons in configurational perturbation theory: Applications in electronic spectroscopy. Advances pentacene/C60 bilayers. J Am Chem Soc 132(36):12698–12703. in Chemical Physics: New Methods in Computational Quantum Mechanics,eds 17. Congreve DN, et al. (2013) External quantum efficiency above 100% in a singlet- Prigogine I, Rice SA (John Wiley & Sons, Hoboken, NJ), Vol 93, pp 219–331. exciton-fission-based organic photovoltaic cell. Science 340(6130):334–337. 36. Pierloot K, Dumez B, Widmark P-O, Roos B (1995) Density matrix averaged atomic 18. Eaton SW, et al. (2013) Singlet exciton fission in polycrystalline thin films of a slip- natural orbital (ANO) basis sets for correlated molecular wave functions. Theoret stacked perylenediimide. J Am Chem Soc 135(39):14701–14712. Chim Acta 90(2-3):87–114.

5330 | www.pnas.org/cgi/doi/10.1073/pnas.1422436112 Zirzlmeier et al. Downloaded by guest on September 27, 2021